A mixed crash frequency estimation model for interrupted flow segments

Document Type

Conference Proceeding

Publication Date

1-1-2019

Abstract

This study explores the link between crash frequency and driver behavior on interrupted flow segments. Among the evolving body of research around this issue, the proposed approach is unique because, among other factors, it also takes into account adjacent land use for the crash analysis segments. Driver behavior data collected from GPS sensors are used to derive a variable termed ProJaT (proportion of jerk above threshold) for each of the 68 analysis segments in three arterial corridors in Baton Rouge, LA. A data-driven approach based on the estimation of correlation coefficients between ProJaT and segment crash rate is used to select the threshold used to define ProJaT for analysis segments. To increase the explanatory power of the model, additional variables such as ADT (average daily traffic), traffic signal presence, adjacent population density, and presence of curvature were also used in the crash frequency estimation model. The results of random-effect negative binomial (RENB) model with commercial land use as a random effect suggested that ProJaT is only significantly associated with crash frequency in segments adjacent to non-commercial land use. On segments adjacent to commercial land use, ADT and influence of traffic signals are significantly related to crash frequency. Implications of the results for proactive identification of crash-prone locations through driver behavior data collected using low-cost GPS devices are discussed in the study.

Publication Source (Journal or Book title)

International Conference on Transportation and Development 2019: Smarter and Safer Mobility and Cities - Selected Papers from the International Conference on Transportation and Development 2019

First Page

72

Last Page

83

This document is currently not available here.

Share

COinS